A supervised ontology-aware cell annotation method for single-cell transcriptomic data

bioRxiv (Cold Spring Harbor Laboratory) 2026 Preprint

View at publisher
Authors: Nimish Magre, Ebtisam Alshehri, Fedor Grab, Yerdos A. Ordabayev, Steven A. McCarroll, Mehrtash Babadi, Stephen J. Fleming
DOI: 10.64898/2026.01.13.699356
Cited by: 1
References: 14

Abstract

Many single-cell RNA-seq annotation methods ignore the hierarchical nature of cell type classification. We present a probability propagation strategy that enforces ontological consistency and improves performance when applied to existing models without retraining. Combined with a lightweight logistic regression model trained on 42 million human cells, this yields SOCAM, a fast and interpretable classifier. We also introduce a hop-based F1 score for ontology-aware evaluation. Code and models are available open source.

Fields of Study & Subjects

Single-cell and spatial transcriptomics Cell Image Analysis Techniques Cancer Genomics and Diagnostics

Publication details

Journal / source
bioRxiv (Cold Spring Harbor Laboratory)
Publisher
openRxiv
Publication date
14 January 2026
Language
EN
Document type
Preprint